SU516188A1 - Device for quantizing a random process - Google Patents

Description

the relational function on the conversion interval of the total; V measurement cycles.

According to commands from control unit 1, the first analog latch 2 detects the instantaneous value U (ti) of the random process U (t), the instantaneous value L / (ti) is simultaneously fed to the input of the voltage converter 3 and the input to the subtractor 4. Analog-digital the converter performs one clock cycle tj. converting the instantaneous value of U (ti) to the code Ki. The distribution of the m quantization levels over the voltage conversion range to the code at the beginning of measurements is selected on the basis of coarse a priori data on the type of correlation function on the conversion interval tn The statistical correction block 5 in accordance with the selected quantization level distribution over the conversion range forms the discrete D / Xu difference codes, D / Sao,. . ., D / Hundred neighboring m quantization levels (quanta codes), which are fed to converter 3 to control the keys of its discrete divider. Depending on the result of the first clock t, equilibration from the converter 3 imparts the signal to the statistical correction block 5, which forms the result code of the first conversion U (ti) into the code.

With a delay Tj equal to one cycle / t conversion, the second analogue latch 6 fixes the instantaneous value f7 (ti + Ti) of the slightest process U (t). The value of TI may be equal to one tx of the transformation or more depending on the length of the portion of the measured correlation function. The subtraction unit 4 forms the difference between the shifted instantaneous values of U (ti + and U (ti)) and feeds it to the statistical processing unit 7 at one input of its comparator 8. Together with the comparator 8, the reversing counter 9, the digital-to-analog converter 10, the gating unit I (statistical processing unit 7) the cycle of determining the code / С -; i, evaluating the restoration error Е, is performed on the interval TJ, equal to fb according to the algorithm

, n, + D / S sign (t, -.

where Kl-, 1. , KP, - codes of errors DYAL- :,,,

recovery in the TI interval in n + 1 and n measurement cycles;

D / S - correction step code in the cycle;

D (/ (t1) J (ti TI) - U (ti) is the increment of the random process on the interval TI;

sign - sign functions

The code Kp, the errors of the recovery of the LEP, are rewritten from the statistical correction block 5 to the reversing counter 9 of the statistical processing unit 7. The resulting code Kn-i i, the recovery errors & En + ii at the end of the cycle is recorded from the reversible counter 9 of the statistical block processing in the first register of the block of statistical correction 5.

Synchronous, contact operation of the analog-digital converter 3, the second analogue latch 6, subtraction unit 4, all nodes of statistical processing units 7 and statistical correction 5 for the t cycles of the tr conversion of the instantaneous value U (ti) into the code in the registers of the statistical correction unit 5 and the following codes are formed Kn + ii, errors of EP + II, recovery at intervals t, (everywhere i 1, 2, ..., t), multiples of the tr tr of conversion. At the same time, the second analog latch 6 fixes the shifted instantaneous values of U (ti - + TJ) at the end of i 1, 2, 3 ,. .., m

conversion cycles and monitors the change in the random process in conversion cycles.

Converter 3 regularly signals the balancing process to the statistical correction block 5, by which it forms the code Kg of the result of converting the instantaneous value U (ti) in the form of the rth algebraic sum of the required difference codes A / (rn, (i 1, 2, 3 , ..., m).

In this way, in parallel with the conversion cycles of the instantaneous value U (ti) of the random process U (t), the determination of the error estimate codes (t 1, 2, ..., m) of the recovery error estimates in the interval

convert n random process into code. After L cycles of simultaneous quantization of a random process L / (f) in analog-digital converter 3 and static processing in block 7, the registers of the block of statistical correction 5 will determine m KNI codes (i 1, 2, ..., m) recovery errors , from which the ordinates τ of the correlation function of the random process U (t) can be determined over the transformation interval t.

It can be shown that the magnitude of the differences

KN /, - KNI I Qk characterize the speed

changes in correlation of random

process on t 1, 2,. . ., t ticks convert it to code.

The codes for the indicated differences in the errors of restoration and correction in accordance with these codes for the differences of the initial codes for the differences D / Co are carried out in the block of statistical correction 5.

The correction of the initial distribution of the quantization level differences codes D / Co / by the error difference codes of restorations allows one to reduce the rms quantization error of the random process into the N discrete sampling codes of one of the registers of the statistical correction block 5, the code KN /, the recovery errors for the correlation interval t / 1 of the random process U (t), according to which it is possible to control the quantization range in the analog-digital converter 3.

five

Using the random process quantization method described as one of the adjective primary processing channels for multi-channel digital correlometers will significantly reduce the rms quantization error without increasing the number of levels in the quantization unit and the size of the subsequent calculating and representing digital correlation units. In addition, it makes it possible to analyze the initial part of the correlation function for a longer time.

The upper frequency of the random process being investigated by the described random process quantizer is limited by the speed of the analog latches, and not by the time the random process is converted to code, which significantly reduces the requirements on the latter without increasing the first minimum interval of the random process measured correlation function.

Thus, the proposed random process quantization device allows parallel quantizing a random process and measuring its autocorrelation function on the transformation interval, performing a statistical correction of the quantization levels on the correlation function on the transformation interval, which makes it possible to improve the accuracy of poiicTBa.

Claims (1)

Formula invented and A device for quantizing a random process containing a statistical processing unit, an analog-to-digital converter connected to a control unit, connected to the first and second analog clamps respectively and to the control input of the subtraction unit, the first input of which is connected to the first analog clamp and the input The device is connected to the inputs of analog latches, characterized in that, in order to reduce the root-mean-square error, the device contains a statistical correction unit connected in moves respectively to the output of the analog-digital converter, to the control unit and to the statistical processing unit, and outputs connected to the analog-digital converter whose input is connected to the output of the second analog clamp and one of the 6vTOKa inputs of the statistical processing connected to the other input with the output a subtraction unit, the second input of which is connected to the output of the second analog lock, which is connected to the information input of the analog-digital converter.